Independent Deployment

• In both separated and independent deployment modes, BoostIO has an independent bio_daemon process. It is specified that only one bio_daemon process can exist on each physical node to ensure data consistency.
• Before starting the bio_daemon process on a compute node that does not support data caching, you need to set the cache size in the bio configuration file to 0 as follows:

  bio.mem.size_in_gb = 0
  bio.disk.path =

You need to start the BoostIO process bio_daemon and then the JuiceFS process. During the startup of the JuiceFS process, the BoostIO SDK library is automatically loaded. The bio_daemon process can be started in either of the following methods:

• Systemd

  You are advised to use this method if you want the bio_daemon process to be taken over by the OS and automatically restart for service continuity when a fault or exception occurs.

• Manual startup at the background

  You are advised to use this method if you want to monitor the bio_daemon process to immediately locate and rectify faults or exceptions. This method is suited for application development and function debugging scenarios.